Understanding Glass Breakage in Field-Operated PV Modules: Learnings from Inhomogeneous Mechanical Load Tests

A significant increase in glass breakages in photovoltaic (PV) power plants has been observed in recent years, particularly in glass/glass (G/G) modules with thin (~2 mm) glass and large areas (>2.5 m²). These failures often occur within months of field deployment and affect various mounting configurations, including framed and unframed modules on tracked or fixed sub-structures. Traditional causes, such as severe weather or faulty installation, can largely be excluded. Furthermore, even modules that passed the mechanical load (ML) tests during certification according to IEC 61215 are currently failing in the field. Laboratory tests and field observations reveal distinct fracture patterns, suggesting fundamentally different triggering mechanisms. Field fractures are often characterized by long, singlerunning cracks, originating within the surface, with no clear connection to glass edges. In contrast, laboratory tests under homogeneous load scenarios typically show chaotic crack patterns originating at clamp positions due to stress concentrations. The reduced mechanical stability of thinner glass, particularly with lower surface pre-stress, is assumed to be one key factor, while other factors as for example the influence of more realistic load profiles remain unclear. This study investigates the mechanical failure behavior of G/G and glass/backsheet (G/B) modules under homogeneous and inhomogeneous load profiles and sets them into context to the currently observed breakages in the field. Results show that G/G modules consistently exhibit lower failure loads compared to G/B modules in the respective mounting configuration replicating a 1P tracker setup. However, G/G and G/B modules demonstrate a distinct mechanical response to specific inhomogeneous load profiles, with failure loads converging under increasingly inhomogeneous conditions. Based on the laboratory findings, field fractures are hypothesized to result from sustained, low-intensity loads. Contributing factors may include sagging-induced tensile stress or low-energy impacts, such as stone-strikes during grass mowing. The findings highlight the limitations of current ML tests, which fail to replicate real-world load conditions. Further research is needed to refine laboratory tests, integrate cyclic loads, and investigate defect formation and aging effects. Such future efforts will be essential for identifying root causes and guiding the development of more robust module designs and maintenance practices to mitigate future failures.